• liver fat;
  • abdominal fat;
  • computed tomography;
  • waist circumference


  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References

Objective: To examine the independent associations of abdominal fat (visceral and subcutaneous) and liver fat with all-cause mortality.

Research Methods and Procedures: Participants included 291 men [97 decedents and 194 controls; mean age, 56.4 ± 12.0 (SD) years] who received a computed tomography (CT) examination at the preventive medicine clinic in Dallas, TX, between 1995 and 1999, with a mean mortality follow-up of 2.2 ± 1.3 years. Abdominal fat was determined using contiguous CT images from the L3-L4 to L4-L5 intervertebral space. Liver fat was assessed using the CT-determined liver attenuation value, which is inversely related to liver fat. Logistic regression was used to determine the independent association between the fat depots and all-cause mortality.

Results: During the study, there were 97 deaths. Visceral fat [odds ratio (OR) per SD: 1.83; 95% CI: 1.23 to 2.73], abdominal subcutaneous fat (1.44; 1.02 to 2.03), liver fat (0.64; 0.46 to 0.87), and waist circumference (1.41; 1.01 to 1.98) were significant individual predictors of mortality after controlling for age and length of follow-up. In a model including all three fat measures (subcutaneous, visceral, and liver fat), age, and length of follow-up, only visceral fat (1.93; 1.15 to 3.23) was a significant predictor of mortality.

Discussion: Visceral fat is a strong, independent predictor of all-cause mortality in men.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References

It is well established that abdominal obesity is associated with increased morbidity independently of age, race, and sex (1, 2). Abdominal adiposity as measured by waist circumference is a significant predictor of morbidity (3) and mortality (4) independently of BMI. The association between waist circumference and metabolic risk may be explained, in part, by its strong association with visceral adiposity (5). Visceral fat is a strong, independent predictor of dyslipidemia (6, 7, 8) and insulin resistance (8, 9), and changes in visceral fat are associated with concomitant changes in glucose tolerance and insulin resistance (10). In prospective studies, visceral fat predicts the development of type 2 diabetes (11). Although the association between visceral adiposity and morbidity seems to be well established, to date, no study has examined the association between visceral fat and mortality using direct measures of visceral fat.

Emerging evidence suggests that fat deposition within the liver is another component of fat distribution that is independently linked to dyslipidemia (8) and insulin resistance (12, 13). Although the association of these fat depots with metabolic risk and morbidity (6, 7, 8, 9, 12, 13, 14) is established, their association with mortality is unknown. Evaluation of the independent associations of abdominal (subcutaneous and visceral) and liver fat with mortality would clarify and reinforce primary targets for obesity reduction and, thus, would have direct relevance to the development of public health policy and clinical recommendations. Therefore, we studied the independent associations between visceral, abdominal subcutaneous, and liver fat deposition and mortality in men.

Research Methods and Procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References


Study participants consisted of a subset of 291 men (97 decedents and 194 controls) who were selected from a larger predominantly white (∼98%) cohort who received a computed tomography (CT)1 examination of the abdomen region as part of a preventive medicine diagnostic exam at the Cooper Clinic in Dallas, TX, between 1995 and 1999, and were followed through December 31, 1999. Study participants were from mid- to upper social economic status groups (∼80% were college graduates), and typically were employed in executive or professional positions. The National Death Index was used to identify potential deaths and cause, and official death certificates were obtained and were cross-referenced with the participant's clinical record to confirm a match. To create a control group that was reflective of our study cohort, the control subjects were randomly selected from survivors that were alive at the time of death of the decedent. The mean length of follow-up was 2.2 ± 1.3 years.

All study participants gave their informed written consent before participation in the examination, and the study was reviewed and approved annually by The Cooper Institute Institutional Review Board. All study participants completed a questionnaire on demographic factors, health habits, and medical history and had a CT exam of the abdomen region. Self-reported height and weight were used to calculate BMI.


Axial images of the abdomen region were obtained using an electron beam CT (Imatron; General Electric, Milwaukee, WI) scanner with the subject lying in a supine position (15). Approximately 40 contiguous images (6 mm thickness) were acquired from the distal iliac crest to the caudal region of the heart. Images were obtained using standard procedures and analyzed using specialized image analysis software (Tomovision, Montreal, Canada). Waist circumference was determined from the CT image at the level of L4-L5, which approximates the level of the iliac crest. A contiguous series of five to seven CT images between the L4-L5 and L3-L4 vertebral disc spaces for each subject were analyzed for determination of abdominal subcutaneous and visceral fat. The fat volumes were calculated using a truncated pyramid method as described previously (5). Adipose tissue volumes (liters) were converted to mass units (kilograms) by multiplying the volumes by the assumed constant density for fat (0.92 kg/liter). Adipose tissue areas (centimeters squared) were computed using an attenuation range of −190 to −30 Hounsfield units. Visceral fat was determined by delineating the intra-abdominal cavity at the innermost aspect of the abdominal and oblique wall musculature and the anterior aspect of the vertebral body. Abdominal subcutaneous fat area was defined as the area of adipose tissue between the skin and the outermost aspect of the abdominal muscle wall.

We previously reported the inter-observer reliability error for visceral adipose tissue and abdominal subcutaneous tissue measurement for two observers’ analyses of the same L4-L5 image (n = 40) to be ∼3% and ∼1%, respectively (16).

CT is capable of differentiating tissues on the basis of their attenuation characteristics, which are a function of tissue density and chemical composition. A normal liver is usually denser and, consequently, has a higher attenuation value than the spleen. Therefore, a lower mean liver attenuation value relative to that of the spleen is an indication of fatty infiltration of the liver (17). Liver fat was represented as both the liver attenuation (CTL, Hounsfield units) by itself, as well as the ratio of the liver to spleen (CTS) attenuation values (CTL/CTS) (17, 18). CTL and CTS were calculated using the average attenuation values for two regions of interest within each organ obtained from a single CT image that clearly displayed both the liver and spleen. The regions of interest were consistently placed in the parenchyma of the right lobe of the liver and in a similar region within the spleen, being careful to avoid blood vessels, artifacts, and other areas of inhomogeneity.

Statistical Analysis

Independent sample Student's t tests and χ2 tests were used to determine group differences in subject characteristics. An unmatched case-comparison study with simple random sampling was used to identify the comparison subjects to estimate the independent association between abdominal fat depots and all-cause mortality (19). Logistic regression was used to estimate the odds ratios (ORs) that are expressed per SD unit to facilitate comparisons of ORs between different adipose tissue depots. Age, follow-up time, abdominal fat depots, and liver fat (CTL or CTL/CTS) were entered in the models as continuous variables. Linear, quadratic, and cubic terms were tested for each continuous independent predictor variable using a stepwise approach with a p < 0.1 to enter and a p < 0.05 to remain in the model. Self-reported smoking status (current, past, never), blood pressure (normal, high), cholesterol (normal, high), triglycerides (normal, high), and type 2 diabetes (yes, no) were entered as categorical covariates. The ORs and 95% CIs are reported after adjustment for age and follow-up time (Model 1) and after further adjustment for all three fat measures (visceral fat, abdominal subcutaneous fat, and liver fat; Model 2). The figures were plotted using the fifth percentile of visceral fat (mass, 0.15 kg; area, 50 cm2) as the referent value. All statistical analyses were performed using SAS v8 (SAS Institute, Cary, NC).


  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References

Subject characteristics are shown in Table 1. The decedents were significantly older and had more visceral fat (P < 0.05) than the controls. The decedents also had a significantly higher prevalence of self-reported type 2 diabetes, high blood pressure, and past smoking than the controls (P < 0.05).

Table 1. . Subject characteristics
  • CTL, liver attenuation; CTS, spleen attenuation.

  • *

    Significantly different from controls at p < 0.05.

  • Values are presented as mean (SD).

  • Decedents (n = 73); control (n = 97).

Follow-up time (years)1.8 (1.2)2.9 (1.0)*
Age (years)51.8 (9.3)65.6 (11.4)*
BMI (kg/m2) †,26.8 (3.2)26.4 (4.8)
Waist circumference (cm)94.8 (9.2)96.7 (11.2)
Visceral fat mass (kg)0.54 (0.25)0.68 (0.39)*
Abdominal subcutaneous fat mass (kg)0.75 (0.29)0.73 (0.34)
Visceral fat area (L4–L5) (cm2)141.3 (61.1)174.2 (96.6)*
Abdominal subcutaneous fat area (L4-L5) (cm2)226.1 (86.2)217.4 (96.6)
Mean liver attenuation (HU)58.3 (10.7)55.8 (12.0)
CTL/CTS1.12 (0.19)1.17 (0.23)
High blood pressure (%)14.930.0*
High cholesterol (%)25.826.8
High triglycerides (%)13.416.5
Type 2 diabetes (%)2.111.3*
Current smoker (%)12.412.4
Past smoker (%)24.740.2*

The associations between the various fat depots and mortality are shown in Table 2. After adjustment for age and follow-up time (Model 1), waist circumference, liver fat, abdominal subcutaneous fat, and visceral fat (Figure 1 A and b) were significant predictors of mortality (p < 0.05). In a model including all three fat measures (visceral, abdominal subcutaneous, and liver fat), age, and length of follow-up (Model 2), only visceral fat mass (OR, 1.81; 95% CI, 1.04 to 3.14; p < 0.05) and area (L4-L5; 1.69; 0.99 to 2.89; p = 0.5) were significant predictors of mortality (Table 2; Figure 1C and D). These results remained similar when traumatic deaths (n = 7) were excluded, after controlling for past or present smoking status, BMI, comorbidities (i.e., high blood pressure, high cholesterol, type 2 diabetes, and high triglycerides), or when examining premature mortality in a subgroup of men 40 to 65 years of age (data not shown).

Table 2. . ORs for all-cause mortality
 Model 1Model 2
 βSEOR (95% CI)*pβSEOR (95% CI)*p
  • Model 1: control for age + follow-up time. Model 2: control for age, follow-up time, abdominal subcutaneous fat, visceral fat, and liver fat. OR, odds ratio; SE, standard error; CTL, liver attenuation; CTS, spleen attenuation.

  • *

    The ORs for mortality in the table are expressed per SD for each variable.

  • Cubic transformation used in the analyses.

Visceral fat mass (kg)0.850.291.83 (1.23 to 2.73)0.0030.860.411.81 (1.04 to 3.14)0.04
Subcutaneous fat mass (kg)1.200.581.44 (1.02 to 2.03)0.04−0.020.780.99 (0.63 to 1.58)0.98
Visceral fat (L4–L5) (cm2)5.0 × 10−61.9 × 10−71.76 (1.16 to 2.67)0.0084.7 × 10−62.4 × 10−71.69 (0.99 to 2.89)0.05
Subcutaneous fat (L4–L5) (cm2)0.0030.0021.31 (0.93 to 1.85)0.15−0.0020.0030.85 (0.53 to 1.36)0.49
Waist circumference (cm)1.19 × 10−65.89 × 10−71.41 (1.01 to 1.98)0.04−1.7 × 10−61.53 × 10−70.60 (0.25 to 1.44)0.26
Mean liver (HU)− (0.46 to 0.87)0.005− (0.56 to 1.37)0.55
CTL/CTS−1.370.820.76 (0.54 to 1.05) (0.64 to 2.48)0.50

Figure 1. The OR for mortality with increasing visceral fat mass (A) and area (B) after control for age and follow-up time. The OR for mortality with increasing visceral fat mass (C) and area (D) after control for age, follow-up time, liver fat, and abdominal subcutaneous fat. The fifth percentile of visceral fat (mass, 0.15 kg; area, 50 cm2) was used as the referent value.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References

In this study we showed that visceral, abdominal subcutaneous, and liver fat (i.e., low liver attenuation) were all directly associated with higher risk of mortality in men. However, visceral fat alone independently predicted risk of mortality after adjustment for the other fat measures. Waist circumference was also directly associated with mortality; however, the association did not remain significant after adjustment for visceral and subcutaneous fat. It has been hypothesized that visceral fat would be directly related to mortality risk based on plausible biological mechanisms (7, 8, 20, 21); however, to our knowledge, this is the first study to show an association of visceral fat with mortality where visceral fat was quantified directly by CT measurements.

Our findings suggest that the relationship between visceral fat and mortality is curvilinear. Inspection of Figure 1 reveals that a rather large amount of visceral fat (∼200 cm2 at the L4-L5 level) accumulation is required before observing a substantially elevated risk of mortality. This threshold is greater than the level of visceral fat (∼110 to 130 cm2 at L4-L5) thought to be associated with a disproportionate increase in metabolic risk in moderately aged men (20). It is likely that these observations are consistent. In other words, a longer follow-up period in our study may have revealed that, at a younger age, our sample of older men had levels of visceral fat in the order of 130 cm2 combined with increased metabolic risk factors. It is well established that visceral fat accumulation increases with advancing age (22).

To our knowledge, this is the first study to show that liver fat is directly associated with mortality risk. Stated differently, liver density was inversely associated with mortality, showing that the leaner the liver, the lower the risk of mortality. Although liver fat was not associated with mortality after adjustment for visceral and subcutaneous fat, the unique contribution of liver fat to morbidity is consistently reported. This notion is underscored by clinical observations showing that liver fat is positively associated with insulin resistance (8, 12) and dyslipidemia (7, 8), independently of visceral fat. A definitive measure of liver fat that is appropriate for use in clinical settings is unknown. However, it is noteworthy that, in our study, waist circumference was a significant correlate of liver fat (waist circumference vs. liver fat, r = −0.59, p < 0.001). This confirms previous observations (23) and sets the stage for further study to determine whether waist circumference can be a useful anthropometric surrogate of fatty liver.

It is clear that both visceral and liver fat are associated with numerous metabolic risk factors, which themselves are antecedents for morbidity and mortality (6, 7, 24). Contrary to the originally proposed substrate-based mechanism, often termed the Portal Theory (25), excess release of proinflammatory and prothrombotic factors may be the mechanism that links visceral fat and metabolic risk (26, 27). The mechanisms that link liver fat per se with metabolic risk, in particular insulin resistance, are largely unknown (24). Emerging evidence does strongly suggest, however, that liver fat is related to insulin resistance independently of abdominal obesity (12) and that reductions in liver fat through weight loss (28, 29) or thiazolidinedione treatment (30) are related to a corresponding reduction in insulin resistance.

The findings of this study provide support for the recommendation that visceral fat should be a primary target of therapeutic strategies designed to reduce obesity-related morbidity and mortality. It is encouraging that visceral fat is substantially reduced in response to diet- or exercise-induced weight loss independently of age and sex (10, 31). It is generally reported that weight loss in the order of 10% is associated with a corresponding reduction in visceral fat that approximates 35% (32). Although routine measurement of visceral fat is not yet feasible, it is established that waist circumference is a strong correlate of visceral fat (33) and, more importantly, that reductions in waist circumference are related to corresponding reductions in visceral fat in a dose-response manner (33, 34). Combined with our finding that waist circumference was directly associated with mortality (Figure 1), these observations support the recommendation that waist circumference should be a routine measure in clinical practice to characterize health risk and, of equal importance, to follow the success of strategies designed to reduce obesity and related comorbid conditions.

Limitations of this study include the use of a relatively small opportunistic sample of men who had acquired abdominal CT images at the Cooper Institute. The men in this study were predominantly white and from a middle-to-upper class population. This limits the generalization of the results of our study but should not affect the internal validity. Indeed, the homogeneity of our study group on socioeconomic factors is a benefit because it reduces the likelihood of confounding by these factors. Simple random sampling from our overall cohort to identify comparison subjects resulted in a large age difference between decedents and comparison participants. This is similar to substantial differences between decedents and survivors seen in prospective studies, and, as in prospective studies, this difference is accounted for in the multivariable statistical model used in our primary analyses (19). The follow-up time in our study was relatively short, and many of the decedents had morbid conditions (e.g., hypertension, type 2 diabetes) on study entry (55% of decedents vs. 26% of controls). Thus, some of the deaths could be attributed to pre-existing disease and may represent the conduit by which obesity led to increased mortality. Although we relied on self-reported diagnosis for diabetes, hypercholesterolemia, and hypertension, the accuracy of self-reported diagnoses in this cohort has been previously shown. For example, we showed that, in this cohort, self-reported hypertension had a sensitivity of 98% and a specificity of 99% (35). Nevertheless, adjustment for the differences in the existence of the comorbidities at baseline did not change the association between visceral fat and mortality.

The strengths of this study include the use of a multi-image CT protocol for measurement of abdominal fat that allowed us to determine subcutaneous and visceral fat mass for a large portion of the abdomen, substantially reducing the potential influence of measurement site (e.g., L4-L5 or L3-L4 intervertebral sites) on our estimate of visceral fat. It is noteworthy, however, that the risk of mortality associated with visceral fat mass and area at the L4-L5 level was not different (Table 2).

Visceral fat is a significant predictor of mortality in men after adjustment for age, follow-up time, subcutaneous fat, and liver fat. These findings underscore the importance of setting abdominal obesity as a primary target for obesity reduction and, consequently, the need to educate health care practitioners on the importance of routine measurement of visceral fat. Although we measured visceral fat directly by CT scans in this study, waist circumference provides a reasonable approximation of visceral fat in clinical settings.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References

The authors thank Elisa Priest for work related to the management of the CT data. This research was supported, in part, by research grants from the NIH to S.N.B. (AG06945) and M.Z.N. and Michael J. LaMonte (HL62508) and from the Canadian Institutes of Health Research to R.R. (MT13448).

  • 1

    Nonstandard abbreviations: CT, computed tomography; CTL, liver attenuation; CTS, spleen attenuation; OR, odds ratio.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgement
  8. References
  • 1
    Okosun, I. S., Liao, Y., Rotimi, C. N., Prewitt, T. E., Cooper, RS. (2000) Abdominal adiposity and clustering of multiple metabolic syndrome in white, black and Hispanic Americans. Ann Epidemiol. 10: 263270.
  • 2
    Nicklas, B. J., Penninx, B. W., Cesari, M., et al (2004) Association of visceral adipose tissue with incident myocardial infarction in older men and women: the health, aging and body composition study. Am J Epidemiol. 160: 741749.
  • 3
    Janssen, I., Katzmarzyk, P. T., Ross, R. (2004) Waist circumference and not body mass index explains obesity-related health risk. Am J Clin Nutr. 79: 379384.
  • 4
    Bigaard, J., Tjonneland, A., Thomsen, B. L., Overvad, K., Heitmann, B. L., Sorensen, TI. (2003) Waist circumference BMI, smoking, and mortality in middle-aged men and women. Obes Res. 11: 895903.
  • 5
    Ross, R., Leger, L., Morris, D., de Guise, J., Guardo, R. (1992) Quantification of adipose tissue by MRI: relationship with anthropometric variables. J Appl Physiol. 72: 787795.
  • 6
    Despres, J. P., Moorjani, S., Lupien, P. J., Tremblay, A., Nadeau, A., Bouchard, C. (1990) Regional distribution of body fat, plasma lipoproteins, and cardiovascular disease. Arteriosclerosis. 10: 497511.
  • 7
    Nguyen-Duy, T. B., Nichaman, M. Z., Church, T. S., Blair, S. N., Ross, R. (2003) Visceral fat and liver fat are independent predictors of metabolic risk factors in men. Am J Physiol Endocrinol Metab. 284: E1065E1071.
  • 8
    Banerji, M. A., Buckley, M. C., Chaiken, R. L., Gordon, D., Lebovitz, H. E., Kral, JG. (1995) Liver fat, serum triglycerides and visceral adipose tissue in insulin-sensitive and insulin-resistant black men with NIDDM. Int J Obes Relat Metab Disord. 19: 846850.
  • 9
    Despres, J. P., Lemieux, S., Lamarche, B., et al (1995) The insulin resistance-dyslipidemic syndrome: contribution of visceral obesity and therapeutic implications. Int J Obes Relat Metab Disord. 19: (Suppl 1), S76S86.
  • 10
    Ross, R., Dagnone, D., Jones, P. J., et al (2000) Reduction in obesity and related comorbid conditions after diet-induced weight loss or exercise-induced weight loss in men. A randomized, controlled trial. Ann Intern Med. 133: 92103.
  • 11
    Boyko, E. J., Fujimoto, W. Y., Leonetti, D. L., Newell-Morris, L. (2000) Visceral adiposity and risk of type 2 diabetes: a prospective study among Japanese Americans. Diabetes Care. 23: 465471.
  • 12
    Tiikkainen, M., Tamminen, M., Hakkinen, A. M., et al (2002) Liver-fat accumulation and insulin resistance in obese women with previous gestational diabetes. Obes Res. 10: 859867.
  • 13
    Marchesini, G., Brizi, M., Morselli-Labate, A. M., et al (1999) Association of nonalcoholic fatty liver disease with insulin resistance. Am J Med. 107: 450455.
  • 14
    Kelley, D. E., Thaete, F. L., Troost, F., Huwe, T., Goodpaster, BH. (2000) Subdivisions of subcutaneous abdominal adipose tissue and insulin resistance. Am J Physiol Endocrinol Metab. 278: E941E948.
  • 15
    Rich, S., McLaughlin, VV. (2002) Detection of subclinical cardiovascular disease: the emerging role of electron beam computed tomography. Prev Med. 34: 110.
  • 16
    Lee, S., Janssen, I., Ross, R. (2004) Interindividual variation in abdominal subcutaneous and visceral adipose tissue: influence of measurement site. J Appl Physiol. 97: 948954.
  • 17
    Piekarski, J., Goldberg, H. I., Royal, S. A., Axel, L., Moss, AA. (1980) Difference between liver and spleen CT numbers in the normal adult: its usefulness in predicting the presence of diffuse liver disease. Radiology 137: 727729.
  • 18
    Ricci, C., Longo, R., Gioulis, E., et al (1997) Noninvasive in vivo quantitative assessment of fat content in human liver. J Hepatol. 27: 108113.
  • 19
    Prentice, R. L., Pyke, R. (1979) Logistic disease incidence models and case-control studies. Biometrika. 66: 403411.
  • 20
    Despres, J. P., Lamarche, B. (1993) Effects of diet and physical activity on adiposity and body fat distribution: implications for the prevention of cardiovascular disease. Nutr Res Rev. 6: 137159.
  • 21
    Matsuzawa, Y., Funahashi, T., Nakamura, T. (1999) Molecular mechanism of metabolic syndrome X: contribution of adipocytokines adipocyte-derived bioactive substances. Ann N Y Acad Sci. 892: 146154.
  • 22
    Han, T. S., Seidell, J. C., Currall, J. E., Morrison, C. E., Deurenberg, P., Lean, ME. (1997) The influences of height and age on waist circumference as an index of adiposity in adults. Int J Obes Relat Metab Disord. 21: 8389.
  • 23
    Scheen, A. J., Luyckx, FH. (2002) Obesity and liver disease. Best Pract Res Clin Endocrinol Metab. 16: 703716.
  • 24
    Yki-Jarvinen, H., Westerbacka, J. (2005) The fatty liver and insulin resistance. Curr Mol Med. 5: 287295.
  • 25
    Björntorp, P. (1990) “Portal” adipose tissue as a generator of risk factors for cardiovascular disease and diabetes. Arteriosclerosis. 10: 493496.
  • 26
    Mohamed-Ali, V., Pinkney, J. H., Coppack, SW. (1998) Adipose tissue as an endocrine and paracrine organ. Int J Obes Relat Metab Disord. 22: 11451158.
  • 27
    Wajchenberg, BL. (2000) Subcutaneous and visceral adipose tissue: their relation to the metabolic syndrome. Endocr Rev. 21: 697738.
  • 28
    Hickman, I. J., Jonsson, J. R., Prins, J. B., et al (2004) Modest weight loss and physical activity in overweight patients with chronic liver disease results in sustained improvements in alanine aminotransferase, fasting insulin, and quality of life. Gut. 53: 413419.
  • 29
    Tiikkainen, M., Bergholm, R., Vehkavaara, S., et al (2003) Effects of identical weight loss on body composition and features of insulin resistance in obese women with high and low liver fat content. Diabetes. 52: 701707.
  • 30
    Bajaj, M., Suraamornkul, S., Piper, P., et al (2004) Decreased plasma adiponectin concentrations are closely related to hepatic fat content and hepatic insulin resistance in pioglitazone-treated type 2 diabetic patients. J Clin Endocrinol Metab. 89: 200206.
  • 31
    Ross, R., Janssen, I., Dawson, J., et al (2004) Exercise-induced reduction in obesity and insulin resistance in women: a randomized controlled trial. Obes Res. 12: 789798.
  • 32
    Janssen, I., Ross, R. (1999) Effects of sex on the change in visceral, subcutaneous adipose tissue and skeletal muscle in response to weight loss. Int J Obes Relat Metab Disord. 23: 10351046.
  • 33
    Ross, R., Rissanen, J., Hudson, R. (1996) Sensitivity associated with the identification of visceral adipose tissue levels using waist circumference in men and women: effects of weight loss. Int J Obes Relat Metab Disord. 20: 533538.
  • 34
    Pare, A., Dumont, M., Lemieux, I., et al (2001) Is the relationship between adipose tissue and waist girth altered by weight loss in obese men? Obes Res. 9: 526534.
  • 35
    Blair, S. N., Goodyear, N. N., Gibbons, L. W., Cooper, KH. (1984) Physical fitness and incidence of hypertension in healthy normotensive men and women. JAMA. 252: 487490.